Visual Press Bias in a Multi-Party Electoral Context

VISUAL PRESS BIAS IN A MULTI-PARTY ELECTORAL CONTEXT

EMBARGOED UNTIL 6PM WEDNESDAY 17 FEBRUARY 2016

CLAIRE ROBINSON

MASSEY UNIVERSITY

NEW ZEALAND

17 February 2016

[email protected]

Clairerobinsonnz.com

@spinprofessor designdemocracy.ac.nz

Claire Robinson Visual Bias in a Multi-party Electoral Context 1 Introduction

News value and balance make for uncomfortable partners in political news reporting. It is the objectivity norm, an intrinsic element of informational journalism (Schudson 2001), that leads to the expectation that the news media should present a neutral and unbiased account of all sides of a political conflict or issue. The requirement for balance becomes particularly salient in elections, when many citizens look to the news media to assist them arrive at, confirm or change a voting decision. The news media has a consequent responsibility to voters, political parties and candidates to present as balanced and fair a picture of the electoral landscape as possible. At a practical level, however, a tension exists between the requirement for balance and the news values or factors that determine what will or will not be firstly selected for investigation and secondly published as news. News values are a system of criteria central to the journalistic decision-making process, and are usually based upon the media’s expert knowledge of the types of stories their audiences are interested in. In the context of political news, factors like incumbency, lead in opinion polls and likelihood of re-election are valued as newsworthy and consequently drive the most coverage (Caple and Bednarek 2013; Green-Pedersen et al. 2015; Hopmann et al. 2012a, b; van Dalen 2012; Waldman and Devitt 1998). This tends to benefit incumbents, an advantage confirmed in studies in a number of different countries which have found that government parties and their leaders get more media coverage than opposition parties and leaders, an effect strengthened whenan incumbent’s re-election chances are strong (Green-Pedersen et al. 2015; Hopmann et al. 2012a; Tresch 2009).

This slant in media attention towards the incumbent on the basis of their newsworthiness is referred to as a structural bias (Hopmann et al. 2012a; van Dalen 2012). This is semantically distinct from partisan bias, which is an individual journalist or news organisation’s ideological preference for one party or leader over another. In reality, justification of bias on the basis of news value rather than partisanship is unlikely to satisfy opposition political parties and their supporters at election times, however, should they find their ability to participate in a fair contest compromised and their visibility minimised by the news media’s effective facilitation of an incumbent’s campaign. Structural bias may cause particular difficulties for smaller parties in multi-party electoral contests or multi-candidate contests like US primaries. Multi-party or candidate electoral contests are contexts where it can reasonably be argued that smaller parties or less high profile candidates have alternative perspectives that are deserving of media coverage (Perloff 2014). Yet

Claire Robinson Visual Bias in a Multi-party Electoral Context 2 because they are in a less powerful position they tend to have fewer resources to promote their platforms and find it hard to attract the same levels of media attention as the more established parties or high profile candidates. The ‘mere exposure’ psychological effect, by which people’s preference for a stimulus is increased by repeated exposure to it (Kahneman 2011), helps us understand why those parties and candidates who are out of sight and not top of mind when voters are making decisions are at a distinct disadvantage compared to those who are highly visible in the news (Iyengar and Kinder 1987, Hopmann et al. 2011, Jenkins 1999).

In political systems with two dominating political parties or candidates the most common definition of balance is ‘equal treatment’ in news coverage of both. For media reports to be considered unbiased it is generally expected that over a period of time there should be as many overtly opinionated statements about one side as the other (D’Alessio and Allen 2000; Jenkins 1999). Equal treatment is harder to achieve in a multi-party electoral context. As Hopmann et al. (2012a) point out:

Requiring equal amounts of media coverage for each political party would ignore the differences between parties that have different electoral sizes and play different roles within a political system. Allocating equal amounts of media coverage to all parties ignores these differences. Hence the question is how political balance in news coverage can be defined in countries with complex party systems (p. 4).

Until now this question has been discussed primarily in European electoral contexts (see van Dalen 2012, Green-Pedersen et al. 2015, Hopmann et al. 2011, 2012a, b, Tresch 2009, for example). If we are to have a have a more comprehensive understanding of this issue it is important to add perspectives from other multi- party electoral contexts.

This study makes a contribution from the antipodes, using the 2014 New Zealand general election as a case study. New Zealand replaced its First-Past-the-Post (FPP) electoral system with a Mixed-Member-Proportional (MMP) system in 1996 to promote greater fairness between political parties and provide more effective representation of minority parties. Until 1996 the FPP electoral system had been characterised by single party governments led by one of two major parties (Labour or National). Small parties stood candidates in electorates, with minimal success.In the post-1996 MMP environment the two major parties compete to be the dominant partner in multi-party coalition governments, and a plethora of smaller political parties from all across the political spectrum (referred to in New Zealand as minor

Claire Robinson Visual Bias in a Multi-party Electoral Context 3 parties) jostle for third party or coalition partner status. On average 28.8% of New Zealand voters have cast their party vote (the determinant of party proportionality) for a minor party each election since the introduction of MMP1. While the financial costs of entry for new and emerging political parties is low, minor parties receive only a small proportion of the state funding allocated for election broadcasting relative to the two major parties, and consequently rely heavily on the news media to cover their election messages. With data from the New Zealand Election Study showing that over 50% of some party voters in New Zealand wait until the election campaign to make their voting decision (see Chart 1), and advance voting becoming increasingly common2, the consequence of media bias in the campaign period, even if it might only be ‘structural’, is significant.

Chart 1: Average Time of Vote Decision over 5 elections: 1999-2011

0 A long 'me ago In elec'on year but before elec'on campaign During the elec'on campaign Maori 42.6 18.6 38.8 Labour 40.7 18.7 40.7 Na'onal 40.4 22.3 37.4 NZ First 27.4 15.1 57.5 Greens 26.3 16.7 56.9 ACT 20.6 18.4 61

Major parties: National and Labour, Minor parties: Maori, NZ First, Greens, Act

Source: New Zealand Election Study data retrieved from http://www.elections.org.nz

Visibility

Not only is this study different because it is from the antipodes, but it is specifically concerned with visual bias in press coverage of political leaders. Despite it being over 30 years since Doris Graber (1984) argued for greater study of news visuals, the visual dimension is still excluded from most studies of media bias (Caple and Bednarek 2013, Groeling 2013). When visibility is studied in the context of bias

Claire Robinson Visual Bias in a Multi-party Electoral Context 4 research, it is more commonly measured in words than in pictures, including the length of TV sound bites or mentions in sentences and the order of news stories. Because visual images are polysemic they are harder to measure and assign agreed meaning to than words. But this does not render them less valid as objects of study. Indeed, to study visibility without studying “that which can be seen” (Thompson 2011:56) is to ignore a critical carrier of political meaning.

Still portraits of political leaders were originally introduced to newspapers to illustrate stories when the word was considered to be the primary container of ‘objective’ political meaning. Today, supported by and supporting the increased personalization of politics, photographic images frequently accompany a written story about a political leader, in some cases take more column space than the written story, and often lead the story. In this environment the photographic image of a political leader is so much more than a candidate’s portrait. It is now fundamental to the exercise and acceptance of the broader political offering (Robinson 2012a). As news is increasingly consumed online, where photos and headlines drive clicks through to written stories, the leader image only becomes more rather than less important. Studies have confirmed that people, particularly the time poor and those with low interest in politics, use images as heuristics to evaluate leadership characteristics such as competence and trustworthiness, which then carry forward to political judgments, choices and voting decisions (Banducci 2002; Ballew and Todorov 2007; Bean 1992; Bean and Mughan 1989; Capelos 2010; Hall et al. 2009, Leathers and Eaves 2008; Lenz and Lawson 2011; McAllister and Bean 2006; Miller et al. 1986; Olivola and Todorov 2010; Riggio and Riggio 2010).

The potency of leader images lies in the fact that they contain images of people. As social beings hard wired to pay attention to and be interested in other people, particularly their faces, humans are instinctively programmed to interpret nonverbal cues and translate them into meaning (Adler et al. 2007; Remland 2004; Surawski and Ossoff 2006). Researchers have found that the brain can extract conceptual information from photographic images seen for as little as 13 milliseconds (Potter et al. 2014), and that 100ms is sufficient exposure for someone to make a political trait judgment after viewing a still photographic image of a candidate’s face (Ballew and Todorov 2007). Appreciating this, political parties and strategists seek to control the image of political leaders by creating ‘photo-ops’ and staged events for photographers to capture images that portray their leaders in a positive light, supplying pre-selected images with high production values that news media organisations can use, and avoiding news mediation entirely by offering leader

Claire Robinson Visual Bias in a Multi-party Electoral Context 5 images directly to the public through social media channels (Adatto 2008; Marland 2012).

While politicians may increasingly determine the image that will be provided to the public, there are still areas under media control. As Waldman and Devitt have noted, “a days worth of campaigning will produce a variety of photographs of a candidate, some happy, some glum, some determined, some silly” (1998: 309). Newspaper editors have the power to select any one of these images to communicate their paper’s support, indifference or displeasure at something a leader has said or done. One day’s live photo then becomes the next day’s stock image. Newspapers maintain libraries of stored images which they can later edit or assign to another story and recontextualise depending on the message they want to communicate to their audience. By privileging certain leader images over others through treatment and placement the press signals to readers who to pay attention to, and why they are important.

Background

During the 2011 New Zealand election campaign a number of letters to the editor were published in the largest of New Zealand’s daily papers, The New Zealand Herald, alleging biased coverage in favour of the incumbent National Party and its leader, Prime Minister John Key. In reply the Herald’s editor claimed “Over many thousands of words in this campaign, coverage has been equally detailed and exacting of the major parties”3. The New Zealand press is committed to balance in the Statement of Principles of its self-regulatory industry body the New Zealand Press Council. Its first principle directs that “publications should be bound at all times by accuracy, fairness and balance, and should not deliberately mislead or misinform readers by commission or omission. In articles of controversy or disagreement, a fair voice must be given to the opposition view”4.

While the words may have been checked for balance, however, the same could not be said for the visual images. This author’s analysis of the press visual image coverage of the two major party leaders in the 2011 election campaign found evidence confirming a substantial imbalance in favour of the incumbent Prime Minister John Key in number, proportion, tone and front page location of visual images, and away from the then major party opposition leader Phil Goff (Robinson 2012b). Critics in the media and on political blog sites variously asserted that the study was ‘worthless’ and its methodology ‘flawed’, particularly the limitation of the study to four newspapers, the two major party leaders, the coverage of the

Claire Robinson Visual Bias in a Multi-party Electoral Context 6 election campaign period only and the absence of a ‘formula’ to make a meaningful assessment of the influence of a particular photo5.

The 2014 New Zealand general election provided the opportunity to broaden the scope of the study, to test whether the 2011 findings were anomalous, to examine the tensions that exist between news value and balance in political news reporting, to better understand the impact of structural visual image bias on minor party coverage and consider whether a more objective measure than news value is needed to achieve fair and balanced image coverage in a multi-party electoral contest.

Data and Method

The study examined the coverage of 12 leaders of nine political parties competing for news media attention in the 2014 New Zealand general election. This included the leaders of the two major parties and ten leaders of seven minor parties (three minor parties had co-leaders). Minor parties are defined in this study as parties reaching above 1% support in public opinion polls, and/or likely to be in parliament after the election by either crossing the 5% party vote threshold or winningan electorate seat. The dataset for analysis included every photographic image of the 12 leaders published in six newspapers over the 262 days between Wednesday 1 January 2014 to Friday 19 September 2014. The period for analysis was further divided into the pre-campaign period from 1 January to 20 August (writ day), and the official campaign period from 21 August to 19 September (the daybefore election day).

To ensure a representative geographic spread across the country the images were sourced from the daily newspapers in New Zealand’s three largest metropolitan areas (The New Zealand Herald/Weekend Herald in Auckland, The Dominion Post in Wellington and The Press in Christchurch), a smaller regional paper – The Southland Times, and the two national Sunday (weekly) editions – The Herald on Sunday and The Sunday Star Times. As Chart 2 shows, in the third quarter of 2014 (when the election was held) the average reading audience for the combined daily newspapers included in this study was 887,000, which is approximately 20% of the population of New Zealand (4.5m). Newspapers are still an important source of information to New Zealand voters, with two thirds reporting they use newspapers to keep up with election news6.

Claire Robinson Visual Bias in a Multi-party Electoral Context 7 Chart 2: Third Quarter 2014 newspaper circulation and readership

144157 Average Net Circulation per issue

451000 New Zealand Herald (daily) Average Readership per issue

118456

Sunday Star Times (weekly) 367000

99191

Herald on Sunday (weekly) 329000

70211

190000 Dominion Post (daily)

65379

183000 Press (daily)

23742

Southland 63000 Times (daily)

Sourced from: Nielsen Consumer and Media insights. Newspapers Year on Year Comparative Results. http://nz.nielsen.com/products/documents/NewspaperToplinesQ42010-Q42011.pdf; http://www. newsworksnz.co.nz/titles/the-southland-times; http://newspaper.abc.org.nz/audit.html?org=np a&publicationid=%25&mode=embargo&npa_admin=1&publicationtype=19&memberid=%25&ty pe=%25; Retrieved 11 February, 2102

Photographs containing more than one leader were counted separately for each leader and treated as two or more separate images in the analysis. In total this amounted to 1539 leader appearances over the entire year. The total data set included 98 images in advertisements: 55 leader images contained in third party advertisements (paid for by Family First and the New Zealand Aged Care Association), 15 leader images in National Party funded advertisements and 28 leader images in Conservative Party advertisements. Party placed and third party ads magnify the exposure the press has given to that leader through their editorial images, but cannot be considered as evidence of editorial press bias per se. Accordingly, in this particular study het primary dataset (n=1441), analysis, charts and tables do not include ads unless explicitly identified.

Claire Robinson Visual Bias in a Multi-party Electoral Context 8 The primary method of investigation was content analysis. The images were counted for number and measured for size to determine which leaders the newspapers considered more or less relevant and needing to be attended to by readers. The images were also coded for tone as cues for a paper’s support, indifference or displeasure at something a leader had said or done. Two coders examined each image and agreed on tone assignation. One of the frequent criticisms directed at tonal interpretation of visual meaning by human coders is that the coding is non- replicable, subjective and potentially biased. This may be so, but there is also a great deal of literature on non-verbal communication that points to culturally agreed signs connoting positive, neutral and negative tone which mitigates the speculative nature of visual analysis. Drawing upon this literature (see Bucy and Grabe 2008; Robinson 2012a, Stewart et al. 2009 for references) positive tone in this study was coded to images that included facial expressions such as laughter, smiles, images of bonding, touching and physical interaction. Neutral tone was coded to images that communicated authority: unemotional facial expressions and contemplation. Negative tone was coded to unflattering images, including grumpy and fearful facial expressions, aggressive physical postures, and/or a lack of social engagement.

The study first analysed the image coverage data (number of images, area and tone) of the two major party leaders competing to be Prime Minister and leader of the next government using the ‘equal treatment’ definition of balance expected in political systems with two dominating political parties. This data was compared to the data collected in Robinson’s (2012b) study of the 2011 New Zealand election coverage of the two major party leaders.

The data for all 12 party leaders was then compared to three statistical markers of party proportionality

(i) share of party vote won in 2011 election,

(ii) public opinion poll data for February/March 2014, averaged from the two major newspaper polls – the February Ipsos Fairfax poll and the March Herald Digipoll7, and

(iii) the actual 20 September 2014 election results8.

The aim was to assess who gained the most advantage when proportionality, rather than equivalence, was the measure of balanced coverage.

To determine who gained the most advantage from a structural bias towards the newsworthy, the study analysed coverage data in relation to three news factors

Claire Robinson Visual Bias in a Multi-party Electoral Context 9 that commonly drive campaign coverage drawn from three Danish media balance studies (Green-Pedersen et al. 2015; Hopmann et al. 2011; Hopmann et al. 2012b): incumbency, relevance and coalition potential. Denmark and New Zealand are similarly sized democratic nations with proportional electoral systems dominated by multi-party coalition governments. It was assumed that if there is a structural bias effect it would be found across multi-party electoral contexts. These Danish studies suggested that:

(i) the incumbent leader would receive the most news attention by virtue of occupying the position of most power and influence;

(ii) their incumbency advantage would be shared with the parties that support the government and this would be larger than for the opposition party leaders that would have supported an alternative government;

(iii) the gap in media appearances between incumbent party leaders and challengers would reduce during the election campaign as challengers became more relevant to the contest;

(iv) parties with coalition potential (parties needed to form a government) would be more successful in influencing the media agenda than parties with blackmailing potential (parties needed by governments for policy support).

A defining feature of the 2014 campaign was the number of minor party co-leaders (6), with each pairing featuring a male and female co-leader. International studies suggest that political media coverage often suffers from a gender bias with female politicians having a harder time making the news than their male counterparts (Tresch 2009). This study also seeks to determine whether the female co-leaders were treated differently to their male counterparts in the 2014 visual image coverage. This is a particularly interesting question to ask in New Zealand which has had two female Prime Ministers in the past 15 years, and considers itself supportive of female political participation. If the bias is structural it should be expected that co-leaders are similarly advantaged or disadvantaged in image coverage. If the bias is gender-based, then it might be expected that any imbalance favours the male co-leader.

Five of the 12 party leaders were Maori. Although there are fewer international studies about ethnic bias in media representation than gender bias (Zurbriggen and Sherman 2010), research in the US has found that the images of minorities tend to be more negative than that of dominant ethnic groups (Voorhees et al. 2007).

Claire Robinson Visual Bias in a Multi-party Electoral Context 10 On the other hand race has not been found to be an issue when it has come to media coverage of African American President Obama’s candidacies (Perloff 2014). Is there evidence of more negative visual coverage of Maori leaders than their non- Maori counterparts? Two of the co-leaders were female and Maori. Did they suffer a double disadvantage in their visual image coverage?

FINDINGS

Equivalence

Moving into the 2014 New Zealand general election year the National Party was the major party in the incumbent coalition government, having won 47.31% of the party vote in the 2011 general election. The two major newspaper opinion polls in February/March 2014 had the National Party sitting at an average of 50.1% support, with a 19 point opinion poll lead over the major opposition Labour Party (30.65%). Incumbent Prime Minister and National Party leader John Key was also preferred Prime Minister with an over 50 point lead over Labour leader David Cunliffe.

The newspaper with the greatest coverage of the major party leaders was the New Zealand Herald/Weekend Herald, publishing 318 (36.5%) of the 872 Key/Cunliffe election year images included in this study (Table 1 and Chart 3). The newspaper with the least coverage of the major leaders was the Southland Times, responsible for 33 (4%) of the major party leader images in this study. The Southland Times was also the least balanced in its coverage of Key and Cunliffe, with 72.7% of its coverage biased towards Key’s image and a 45.4 percentage point gap between Key and Cunliffe. This was followed by the Dominion Post (Wellington) and the Press (Christchurch). The Auckland based Herald and the two Sunday papers were the most balanced in terms of numerical proportion (Chart 3). Even then there was still a significant 18.8 percentage point gap between Key and Cunliffe in those papers over the full year.

Table 1: Individual newsaper coverage Key vs Cunliffe Full year 2014

Newspaper John Key No (%) David Cunliffe No (%) Total (%) NZ Herald/Weekend Herald 189 (59.4) 129 (40.6) 318 (36.4) The Press 129 (63.2) 75 (36.8) 204 (23.4) Dominion Post 127 (64.5) 70 (35.5) 197 (22.6) Sunday Star Times 43 (58.9) 30 (41.1) 73 (8.4) Herald on Sunday 27 (57.4) 20 (42.6) 47 (5.4) Southland Times 24 (72.7) 9 (27.3) 33 (3.8) 539 333 872 (100)

Claire Robinson Visual Bias in a Multi-party Electoral Context 11 Chart 3: Newspaper proportion of major leader coverage

100%

90%

80%

70%

60%

50% David Cunliﬀe

John Key 40%

30%

20%

10%

0% Southland Times Dominion Post The Press NZ Herald/Weekend Sunday Star Times Herald on Sunday Herald

Over the whole year John Key’s image represented 62% of the number of major party leader images and 61% of area. David Cunliffe’s image represented 38% of number and 39% of area (see Table 2). In the pre-campaign period there was a 20 percentage point difference in Key’s favour in number and 22 point difference in area. Contrary to the expectation that the gap in media appearances between the incumbent leader and the challenger would reduce during the election campaign, Key’s numerical coverage increased in the campaign to 32 points higher than Cunliffe’s, while his area coverage remained at 22 points higher. This numerical point difference was higher than the 26.6 point difference between Key and challenger Phil Goff in 2011 election campaign press coverage. In that coverage Key received 63.3% of the number of images and area to Goff’s 36.7% (Robinson 2012b).

Table 2: Major Party Leader Coverage 2014

Pre-Campaign N0 (%) Campaign No (%) TOTAL (%) John Key % Number 353 (60) 186 (66) 539 (62) David Cunliffe % Number 238 (40) 95 (34) 333 (38) John Key % Area 4611671 (61) 2563655 (61) 7175326 (61) David Cunliffe % Area 2950511 (39) 1633045 (39) 4583556 (39)

Claire Robinson Visual Bias in a Multi-party Electoral Context 12 Key also dominated front page coverage throughout both the pre-campaign period and during the campaign (Table 3). During the 2014 campaign Cunliffe had proportionally more front page exposure than his 2011 predecessor. However, this was proporionally less front page coverage than Cunliffe received in the pre- campaign period.

Chart 4 shows the gap between the numerical coverage of the two major leaders widening over the year. The July spike in favour of Cunliffe was due to a number of multi-page spreads in many of the newspapers introducing readers to the new Labour leader, including large portraits of him and his family and pictures taken out of his family photograph albums.

Table 3: Front page 2011 and 2014: number (proportion)

2011 Campaign 2014 Pre-campaign 2014 Campaign 6 4 papers 6 papers papers John Key 23 (69.7%) 20 (62.5%) 42 (65.6%) Phil Goff (2011) David Cunliffe (2014) 10 (30.3%) 12 (37.5%) 22 (34.4%)

Chart 4: Key/Cunliffe Numbers by month

140

120

100

Key

Cunliﬀe 60

0 January February March April May June July August September

Claire Robinson Visual Bias in a Multi-party Electoral Context 13 Moving into the 2014 campaign, Table 4 and Chart 5 show that in the week leading up to the campaign the two major leaders received parity of image numbers, but for the next two weeks the difference in coverage between the two leaders grew to 53 - 58 percentage points. In weeks 0 and 1 Key had 84 pictures published of him in the papers — an average of 6 a day —Cunliffe had 24 photos, not even 2 a day.The percentage point difference between the two leaders dropped to 14 points in the second to last week of the campaign, only to rise again in the last week to 34 points.

Table 4: Proportion of number during 2014 election campaign, two major party leaders

Number WEEK 0-1 WEEK 0 WEEK 1 WEEK 2 WEEK 3 WEEK 4 9-15 August 16-22 August 23 -29 Aug 30 August - 6-12 Sept 13-19 Sept 5 Sept Key 15 (50%) 39 (76.5%) 45 (79%) 34 (58%) 27 (57%) 70 (67%) Cunliffe 15 (50% 12 (23.5%) 12 (21%) 25 (42%) 20 (43%) 35 (33%)

The steep rises in Key’s campaign image coverage correspond to two newsworthy events that focused the media’s attention on Key specifically. The first related to the 13 August launch of the book Dirty Politics (Hager 2014), centred on hacked emails between a right wing blogger and a political staff member in the Prime Minister’s Office. The mainstream news media took umbrage at the book’s revelation that the blogger had been assisted to gain access to declassified security documents that had been refused to mainstream journalists. The book’s publication led to the release of more emails by the source of the hacked emails, which appeared to implicate the then Minister of Justice in an earlier smear campaign against the then Director of the Serious Fraud Office. This spike in campaign coverage for Key was similar to a spike in coverage for Key found in the 2011 campaign covering ten days of media attention over an incident dubbed ‘Teagate’. At its core this was a disagreement between the Prime Minister and the news media about the boundaries between what is public and private political information when a story has been set up as a photo-op devised for maximum media exposure. During the Teagate phase of the 2011 campaign John Key’s number coverage rose to 71% in both number and area to Phil Goff’s 29%.

Claire Robinson Visual Bias in a Multi-party Electoral Context 14 Chart 5: Actual numbers of images during 2014 campaign, all party leaders

60 Key

Cunliﬀe

50 Norman

Turei

Peters 40 Dunne

Flavell

30 Whyte

Craig

Harawira 20 Harre

0 WEEK 0-‐1 9-‐15 August WEEK 0 16-‐22 August WEEK ONE 23 -‐29 WEEK 2 30 August -‐ 5 WEEK 3 6-‐12 Sept WEEK 4 13-‐19 Sept August Sept

Key’s 2014 image coverage levels returned to almost pre-campaign levels following his decision to accept the resignation of the Minister of Justice on 31 August 2014 pending an inquiry into the allegations to clear her name, only to rise again in the last week of the campaign following more intense media focus on Key about an event dubbed ‘Moment of Truth’. On 15 September 2014 German-Finnish internet tycoon and New Zealand resident Kim Dotcom, who had founded the Internet Party earlier in the year as part of his personal crusade to remove John Key from office, held an Auckland Town Hall event featuring Pulitzer prize winning US journalist Glenn Greenwald, NSA whistleblower Edward Snowden and Julian Assange (the latter two screened live from Moscow and the Equadorian Embassy in London). Dotcom claimed Key had struck a deal with the US authorities to grant him New Zealand residence in order that he could then be extradited to the US to face piracy charges. It was Dotcom who branded the event the ‘Moment of Truth’ and promised to reveal informationthat would inflict as much political damage as possible on the Prime Minister in the run-up to the election. The event attracted a lot of media attention, but in the end it was Dotcom that suffered the greater media backlash after providing no new damning information on Key.

Claire Robinson Visual Bias in a Multi-party Electoral Context 15 Not only did these events see Key’s image coverage rise in number, but his negative image coverage rose significantly during the Dirty Politics phase. Table 5shows little overall difference between Key’s tonal coverage between the pre-campaign and campaign periods. However, when the campaign is separated into the high newsworthy phases described above, Key’s negative tonal coverage rose to a 46% peak during the Dirty Politics phase of the 2014 campaign (Table 6). For the rest of the campaign his positive tonal coverage was at a higher level than it had been throughout the pre-campaign period, and he appeared to experience no negative tonal coverage from the ‘Moment of Truth.’

Labour leader David Cunliffe scored higher than John Key in positive/neutral coverage. But that finding has to be tempered. While a greater proportion of Key’s campaign images in 2014 were negative compared to his 2011 campaign coverage, this treatment appears to have been confined to the Dirty Politics phase of the 2014 campaign, and the number of positive Key images still outnumbered the number of positive Cunliffe images at all times of the election campaign (see Chart 6). When National’s campaign newspaper advertisements featuring a positive image of Key are factored in to the dataset, Key’s positive exposure increases even more (see dotted line in Chart 6).

Table 5: Tone 2011 and 2014, major party leaders

2011 Campaign 2014 Pre-cam- 2014 Campaign (6 (4 papers) paign (6 papers) papers) John Key Positive & Neutral 79.7 69.2 67.6 John Key Negative 20.3 30.8 32.4

Phil Goff/David Cunliffe Positive & Neutral 82.5 74.4 80.6 Phil Goff/David Cunliffe Negative 17.5 25.6 19.4

Table 6: Tone 2014 Campaign phases, major party leaders

Dirty Politics Mid campaign Moment of Truth 21-31 Aug 1-15 Sept 16-19 Sept John Key Positive & Neutral 53.8 74 74.3 John Key Negative 46.2 26 25.7

David Cunliffe Positive & Neutral 87.5 76.4 82.6 David Cunliffe Negative 12.5 23.6 17.4

Claire Robinson Visual Bias in a Multi-party Electoral Context 16 Chart 6: Key and Cunliffe: Full Year Tone

Proportion

When all 12 leaders are included in the data set (Table 7), Key’s image proportion reduces to 36.5% of the number and 37.5% of the area occupied by leader images. Cunliffe’s image proportion reduces to 23% of full year number and 24.2% of full year area. Key’s proportion is around 10 percentage points lower than National’s 2011 election result, 13 points lower than its Feb/March 2014 party opinion poll standing and 9.5 points lower than its actual September 2014 election results. Cunliffe’s image coveragewas also 3-4 points less than Labour’s 2011 27.48% party vote; about 6-7 points lower than Labour’s Feb/March 2014 party opinion poll standing, and 1-2 points less than Labour’s September 2014 25.13% party vote.

Interestingly, as Chart 7a and 7b show, if the minor party leaders are treated as a bloc, for most of election year the combined third party leader image number and area was higher than the number for Cunliffe, and often surpassed the image number and area for Key. By the last week of the campaign the combined numerical coverage of the minor leaders combined was 49.1%, which is a substantial amount more than their combined Feb/Mar opinion poll standing of 18.4% and their combined actual party vote of 26.98% on 20 September 2014.

Claire Robinson Visual Bias in a Multi-party Electoral Context 17 Table 7: All party leaders 2014 Party Vote Vote 2014 Party % 47.04 25.13 10.70 8.66 3.97 1.42 1.32 0.69 0.22 0.85 100 (%) 37.9 24.2 4.1 2 7.9 6 6.1 5 1.8 0.8 3 1.2 100 Full Year Full mm area 7175326 4583556 778388 385979 1498272 1140866 1160447 988484 350342 15323 613457 226552 18916992 % 36.4 23.2 7.2 0.9 10.3 6.2 3.2 5 3 0 2.9 1.7 100 Campaign mm area 2563655 1633045 510590 64037 723916 439496 223242 348886 214072 0 203089 124066 7048094 % 38.9 24.9 2.2 2.7 6.5 5.9 7.9 5.4 1.1 0.1 3.5 0.9 100 Pre- campaign mm area 4611671 2950511 267798 321942 774356 701370 937205 639598 136270 15323 410368 102486 11868898 (%) 37.5 23 5.3 2.4 9 8.5 3.7 3.8 1.3 0.3 3.3 1.9 100 Full Year Year Full number 539 333 76 34 130 123 54 55 19 4 47 27 1441 % 39.5 20 7 1.2 11.9 6.4 2.8 4 1.7 0 3.6 1.9 100 Campaign number 186 95 34 6 56 30 13 19 8 0 17 9 473 % 36.5 25 4.3 2.9 7.6 9.6 4 3.7 1.1 0.4 3 1.9 100 Pre- campaign number 353 238 42 28 74 93 41 36 11 4 30 18 968 Feb/March Feb/March 2014 polls 50.1 30.65 11.55 3.6 1.7 0.2 0.65 0.65 0.05 0.85 100 2011 Party 2011 Party % Vote 47.31 27.48 11.06 6.59 2.65 1.08 1.43 1.07 0.6 0.73 100 Leader (National) John Key Cunliffe (Labour) David Norman (Greens) Russel (Greens) Metiria Turei (New Zealand First) Peters Winston (Conservatives) Colin Craig (Internet/Mana) Hone Harawira (Internet/Mana) Laila Harre (Maori) Flavell Ururoa Te (Maori) Turia Tariana (ACT) Jamie Whyte Future) Dunne (United Peter Other TOTAL

Claire Robinson Visual Bias in a Multi-party Electoral Context 18 Chart 7a: Major parties vs all minor parties as a bloc, full year 2014: number

180

160

140

120

100 Na>onal Key

Labour Cunliﬀe

80 Minor Party leaders

0 January February March April May June July August September

Chart 7b: Major parties vs all minor parties as a bloc, full year 2014: area

3000000

2500000

2000000

Na=onal 1500000 Labour

Minor party leaders

1000000

500000

0 January February March April May June July August September

This could be interpreted as evidence of the press taking minor party leader coverage seriously. But it needs to be remembered that the purple line is the aggregation of the coverage of ten minor party leaders, who each had very different coverage stories. The next section examines the leader coverage organised by the news factors identified in the three Danish studies: incumbency, relevance and coalition potential.

Claire Robinson Visual Bias in a Multi-party Electoral Context 19 Incumbency

Green-Pedersen et al. (2015) suggest that the sharing of policy responsibility and need for support for the introduction of new policies typical of minority rule enhances the policy influence of parties belonging to the same bloc as the government, and creates an incumbent coverage bonus for those support parties. In their study of 31,000 political stories in Danish radio news from 1984-2003 they found that government support parties appeared more in the news than opposition parties that supported the alternative government, with incumbent supporters featuring in 12.7 percent of the news, while those who supported the challengers featuring in eight percent.

Did the minor party leaders from the 2011 National-led coalition government receive a similar incumbency advantage in their 2014 visual press coverage? In New Zealand three minor support parties formed part of the 2011-2014 coalition government: the Maori Party (led by co-leaders Te Ururoa Flavell and Tariana Turia), ACT (Jamie Whyte) and United Future (Peter Dunne). Their combined leader coverage in the pre-campaign period totaled 6.4% in number and 5.6% in area, rising slightly during the campaign period to 7.2% in number and 7.6% in area, and averaging out over the full election year at 6.8% in both number and area. While not as high as the 12.7% incumbent support party levels found in the Danish study this coverage, whether measured in pre-campaign, campaign or full year periods, is at a higher level than their combined 2011 party vote of 3.1%, their Feb/March 2014 opinion polling standing of 2.7%, and their combined 2014 party vote of 2.23%.

Individually, however, these minor party leaders were in the bottom five for number and size coverage over the whole year (see Table 7). Having announced her decision in November 2011 to retire at the 2014 election, Maori Party co-leader Tariana Turia appeared to no longer have news value. For the whole of 2014 she was the least covered of all 12 leaders, and during the campaign received no coverage at all. The other minor support leaders fared not much better in the campaign, especially during the Dirty Politics phase. As shown in Table 8, in weeks 0 and 1 when the media’s focus was on Key during this period, Peter Dunne was the only minor support leader to feature, and then it was only in two images, representing 1.4% of week 0 coverage.

Claire Robinson Visual Bias in a Multi-party Electoral Context 20 Table 8: Proportion of number all parties by campaign week

WEEK 0-1 WEEK 0 WEEK 1 WEEK 2 WEEK 3 WEEK 4

9-15 August 16-22 August 23 -29 August 30 August - 5 6-12 Sept 13-19 Sept

Sept Key (National) 25 54.2 59.2 35.4 24.8 40 Whyte (ACT) 1.6 0 0 2 8.3 3.4 Flavell (Maori) 0 0 0 3.1 2.7 1.1 Dunne (United Future) 1.7 1.4 0 4.2 3.6 0.6 Turia (Maori) 0 0 0 0 0 0 Craig (Conservatives) 6.7 2.8 5.3 3.1 8.3 8 Cunliffe (Labour) 25 16.6 15.8 26 18.3 20 Peters (NZFirst) 11.7 5.6 5.3 9.5 18.3 11.4 Norman (Greens) 5 8.3 7.9 9.5 6.4 6.3 Harre (Internet/Mana) 11.7 1.4 3.9 5.2 2.8 4.6 Turei (Greens) 3.3 9.7 0 1 0.9 2.3 Harawira (Internet/Mana) 8.3 0 2.6 1 5.6 2.3 TOTAL 100 100 100 100 100 100

Was the coverage advantage for the minor support parties larger than for the opposition party leaders that would have supported a Labour government in 2014? The opposition parties that would have supported an alternative Labour government included the Greens (led by co-leaders Russel Norman and Metiria Turei), the Internet/Mana Party (led by co-leaders Hone Harawira and Laila Harre) and (possibly) New Zealand First (led by Winston Peters). The first two parties had publicly stated they would support a Labour-led coalition. New Zealand First had supported the 2005-2008 Labour government. It had not publicly said it would support a 2014 Labour government but it had not ruled it out either.

The combined leader coverage for these partiesin the pre-campaign period totaled 22.5% in number and 24.7% in area, rising during the campaign period to 24.9% in number and 26.6% in area, and averaging out over the full election year at 24.2% in number and 25.1% in area. This is a substantial amount more than the minor support party coverage and the 8% challenger coverage found in the Danish study. In fact, it is almost at the same level as the Labour leader’s election year coverage of 22.5% in number and 24.7% in area. Whether measured in pre-campaign, campaign or full year periods, their coverage is between 4 and 9 points higher than their combined 2011 party vote of 18.73%, their Feb/March 2014 opinion polling standing of 15.35%, and their combined 2014 party vote of 20.78%. In the last week of the campaign the minor challenger party leaders received 26.9% of the total number of leader images. This was five times as much coverage as the minor support party leaders, who only received 5.1% of the total number of leader images

Claire Robinson Visual Bias in a Multi-party Electoral Context 21 in that week, and 6.9 percentage points higher than Cunliffe’s 20.5% coverage. The gap between the coverage for the minor challenger parties and the minor support parties can be seen in the Charts 8a and 8b.

What happens when the major party coverage data is joined with the minor party coverage data? Hopmann et al.’s (2011) content analysis of election news coverage spanning five Danish national election campaigns from 1994 to 2007 found that relative to standing in opinion polls the government experienced in all elections an average incumbency bonus of seven points, whereas the opposition experienced a parallel disadvantage of the same amount. Hopmann et al.’s (2011) study also found that the incumbency bonus was larger when the government was expected to win the election. These findings are not replicated in New Zealand in 2014. Despite National being expected to gain the majority party vote and form the next government, the image coverage bonus was not in favour of the National- led coalition. When the minor support parties are joined with the National Party, the combined incumbent parties rise in number from 42.9% in the pre-campaign period to 46.7% in the campaign, but fall from 44.5% in area in the pre-campaign period to 43.9% in the campaign. Over the full year they record a combined 44.3% number and 44.7% area coverage. This is six points less than their combined 2011 party vote, seven points less than their Feb/March 2014 opinion poll standing, and five points less than their 2014 election result. If anything, the combined incumbent parties recorded a 5-7 point penalty, not a bonus.

Relevance

Green-Pedersen et al. (2015) argue that opposition parties become more relevant and therefore newsworthy during election campaigns than in routine times because they are in the contest, running for seats, office and future influence. In their study of 31000 political stories in Danish radio news from 1984-2003 they found that the gap in media appearances between incumbents and challengers reduced during election campaigns. While incumbents experienced a small reduction in coverage during campaigns (-2.30%), challengers increased their appearances by more than 6 percentage points (6.34%) leaving a net difference of 8.64 points.

Claire Robinson Visual Bias in a Multi-party Electoral Context 22 Chart 8a: Major with incumbent and challenger support parties number full year

140

120

100

80 Na>onal Key

Labour Cunliﬀe

Minor Incumbent 60 Minor Challenger

0 January February March April May June July August September

Chart 8b: Major with incumbent and challenger support parties full year area

1800000

1600000

1400000

1200000

1000000 Na>onal Key Labour Cunliﬀe

Minor Incumbent 800000 Minor Challenger

600000

400000

200000

0 January February March April May June July August September

If the comparison is made between the proportion of coverage the challengers received in the pre-campaign and campaign periods, we do not see the same outcome in New Zealand. The combined Labour/challenger parties decreased their proportion of numerical coverage by 0.6% and increased their area coverage by 0.2%. However, as Charts 9a and 9b show, the major increase in challenger coverage

Claire Robinson Visual Bias in a Multi-party Electoral Context 23 occurred not at the beginning of the official campaign period, but 2 months earlier at the beginning of July. If the comparison is made between the January to June coverage and the July to September coverage, there is a ten point increase in favour of challenger coverage, from 41.35% in January to June to 51.9% from July to September in number, and an 18 point increase in area coverage over the same period from 41.35% to 51.9%. While much of this is due to the profiling of Cunliffe in July, by the end of the campaign the combined challenger coverage was higher than for the combined incumbents.

Chart 9a: Incumbents vs challengers number full year 2014

200

180

160

140

120

100 Challenger Incumbent

0 January February March April May June July August September

Chart 9b: Incumbents vs challengers area full year 2014

3000000

2500000

2000000

1500000 Incumbent Challenger

1000000

500000

0 January February March April May June July August September

Claire Robinson Visual Bias in a Multi-party Electoral Context 24 Coalition potential

Hopmann et al.’s 2012 Danish study of 812 news stories and 334 party published press releases in the 2007 Danish national election campaign found that parties with coalition potential (parties needed to form a government) were more newsworthy than small parties with blackmailing potential (that is, needed by governments for policy support). The authors concluded that small parties that are redundant to form a government are largely irrelevant to the news media agenda. There is evidence to support this effect in New Zealand. Despite the early 2014 public opinion polls pointing to a National majority, the media speculated that National would not hold onto that majority moving into the election and would also not be able to rely on its current minor support parties to get it over the line. Instead it would need a larger coalition partner like New Zealand First or the Conservatives to enable it to form the next government. Despite leading the 4th ranked party after the 2011 election, New Zealand First’s Winston Peters was the third most published leader over the 2014 in both number and area. His campaign coverage at 11.9% in number and 10.3% in area was higher than New Zealand First’s final 2014 party vote of 8.66%. Despite not being in parliament the Conservatives’ leader Colin Craig was the fourth most published leader over the whole year in number and 5th in area. His campaign coverage of 6.4% in number and 6.2% in area was over two percentage points higher than the Conservative Party’s final 2014 party vote of 3.97%.

So how did these two potential coalition partners’ coverage sit in relation to the blackmail party coverage? Since 1996 the composition of New Zealand coalition governments has been such that the major party in government has secured enough support from its minor coalition partners to guarantee confidence and supply in the House of Representatives. In 2014 the parties needed to secure confidence and supply were the same parties needed in order to pass legislation — the blackmailing parties can thus be considered to be the Maori, United Future and ACT parties. Together they received 6.8% of the number and area coverage over the full year. This was a substantial amount less than Peters and Craig’s combined coverage of 17.5% of the number of leader images and 13.9% of area coverage over the whole year. Craig’s image alone was more covered than all the minor support party leaders combined over 2014.

How did the coverage for the potential incumbent coalition partners compare to coverage for the coalition partners backing the alternative government? The most likely potential coalition partner for Labour in an alternative government was the Green Party, which had indicated it would support a Labour-led government. As a

Claire Robinson Visual Bias in a Multi-party Electoral Context 25 coalition offering, however, Labour and the Greens did not look competitive in early 2014 opinion polls. With a combined 42.2% standing in the Feb/March polls they were short of a majority and would have needed another coalition partner to be in a position to form the next government. While Labour itself was courting New Zealand First as the other potential coalition option, the news media speculation centred on a Labour/Green/Internet/Mana alternative. As Chart 10 shows, however, this alternative government trio did not attract as much press image coverage as the National/New Zealand First/Conservative grouping. It is a stretch to suggest that these combinations reflected the coalition preferences of the press. But what the chart does show is the image coverage advantage given to certain party leaders when the press values those parties for their potential to form the next government.

Chart 10: Numerical coverage difference between main coalition options 2014

250

200

150

Labour/Greens/InternetMana

100

0 January February March April May June July August September

Gender and ethnicity

Were the female co-leaders treated differently to their male counterparts in visual image coverage? Three parties had female co-leaders (Internet Mana: Laila Harre, Maori: Tariana Turia, and Greens: Metiria Turei). Female leaders thus represented 30% of the minor party leaders and 25% of the total leadership included this study. Their proportion of image coverage did not match this, with only 7% of the number of images, and 8.2% of the area in the pre-campaign period accorded to the female leaders, reducing during the campaign to 5.2% of the number of images, and 5.9%

Claire Robinson Visual Bias in a Multi-party Electoral Context 26 of the area covered. Next to Turia, Green party co-leader Metiria Turei had the least number of images published during the campaign (6) and the least amount of area coverage (64037 column mm).

Chart 11 compares the differences between male and female co-leaders in terms of image number and area over the whole year. Male Green party co-leader Russel Norman received 76 (69%) of the number of Green Party leader images to Metiria Turei’s 34 (31%). His image received 66.9% of the Green party co-leadership image area coverage. The Maori Party’s Te Ururoa Flavell received 19 (82.6%) of the number of images to Tariana Turia’s 4 (17.4%). His image received 96% of the Maori party co-leadership coverage. Only Internet/Mana — two minor parties with two separate leaders that had joined forces specifically to campaign in the 2014 election — had an almost equal distribution of number between the female leader (55) of the Internet Party and the male leader of the Mana Party (54). But Hone Harawira’s image received a greater proportion of area coverage: 54% to Laila Harre’s 46%. Laila Harre was the only non-Maori female co-leader of a minor party. She only took up the party co-leadership in May 2014 at the time the Internet Party was established, and this will have impacted on her coverage statistics. On the other hand, on 21 August 2014 Harawira was involved in a car accident which saw him remove himself from the campaign to recover. This may have impacted on his image coverage as well, dropping from 4% pre campaign to 2.8% during the campaign in number and 7.9% in pre-campaign area to 3.2% campaign area coverage

Chart 11: Female/male co-leader split

Green Party Maori Party Internet/ Mana

Metiria Turei Tariana Turia Laila Harre Russel Norman Te Ururoa Flavell Hone Harawira

Claire Robinson Visual Bias in a Multi-party Electoral Context 27 Maori leadership

Was there evidence of more negative visual coverage of Maori leaders than their non-Maori counterparts? Five of the 12 or 42% of the party leaders in this study were Maori. Despite this high proportion, Maori leaders combined only received 16.7% of the number of leader images and 18.6% of the area coverage over the whole year. Most of the non-Maori proportion is taken up by the two major party leaders. Removing them from the data-set, the remaining five Maori leaders represented 50% of of minor party leaders. They received 42% of the number of minor party leader images and 48% of the minor leader area coverage. Of the five Maori leaders Winston Peters’ image was the most published in number and area. Over the whole year 4 of the 6 least published were leaders who were Maori.

DISCUSSION

This study was concerned with the question of how balance in political news coverage can be defined in countries with complex party systems, with a particular interest in the tension between structural image bias towards the newsworthy and more objective notions of balance such as equivalence and proportionality. The research found no relationship between equitable balance and major party leader coverage. Over the whole year John Key’s image represented 62% of the number of major party leader images and 61% of area. David Cunliffe’s image represented 38% of major party leader images and 39% of area. During the election campaign Key’s numerical coverage was 32 points higher than Cunliffe’s, and his area coverage was 22 points higher. During the Dirty Politics phase of the official election campaign the difference in coverage between the two leaders grew to 53 - 58 percentage points in Key’s favour. Although Key’s proportional negative tonal coverage also rose during this particular phase, positive images of Key outnumbered positive images of Cunliffe all throughout the campaign.

At the beginning of this study it was suggested that structural bias in favour of the newsworthy would negatively impact on minor parties. The evidence shows, however, that it was not minor parties per se that were disadvantaged. As a bloc, minor parties did well in press image coverage in comparison to the two major parties. At the individual party leader level it was a different story. Of the two major parties, the major news factors driving the visual image coverage imbalance in National Party leader John Key’s favour appeared to be incumbency and the likelihood that he would lead the next coalition government. Of the seven minor parties the visual imbalance favoured those that were assessed by the press as being

Claire Robinson Visual Bias in a Multi-party Electoral Context 28 in a position to help National form that next coalition government — principally the leaders of the New Zealand First and Conservative Parties. This pattern of visual imbalance limited the visibility of those that did not fit within that narrow range of news factors. Leaders of other opposition parties and parties that supported the government but were not regarded as ‘kingmaker’ parties did not receive the same levels of favourable coverage.

While the news factors that drove the most coverage were the ones also identified by the three Danish studies: incumbency, relevance and coalition potential, the pattern of New Zealand coverage was different to that found in Denmark. The Danish studies suggested there would be an incumbency bonus for minor parties in the coalition government. While there is clear evidence of a coverage advantage for the incumbent major party, this did not flow on to the minor support parties. If anything, incumbency looks to have disadvantaged the minor parties that were part of the 2011 National-led government, who received the lowest image coverage over the whole year. The Danish studies suggested that challengers became more newsworthy during the election campaign. The New Zealand study found evidence to support this, though not so much in the official election campaign period as dating back three months before the campaign. However, it did not find evidence supporting this for the major challenger. The Labour party leader gained coverage in July but dropped during the election campaign. The strongest link between Danish and New Zealand findings is for coalition potential being a news factor driving coverage. The parties that the press deemed to be of greatest coalition government partnership potential received a significantly higher proportion of the visual image coverage than the parties that had blackmail potential only.

Did ethnicity and gender affect levels of visual coverage? When the differences between the Maori and Green Party female and male co-leaders visual coverage is examined, there is evidence of a bias towards the male co-leader over the female co-leader. This was not found in the case of Internet/Mana pakeha co-leader Laila Harre, however. Being a female party leader on its own was not necessarily a disadvantage then. Nor, given the high levels of coverage enjoyed by Winston Peters was being a Maori leader of a minor party on its own a disadvantage. But being a female Maori co-leader of an incumbent support party or a challenger party not considered by the press to be of coalition potential did appear to be a distinct disadvantage for visual image coverage, particularly in the important election campaign period.

Claire Robinson Visual Bias in a Multi-party Electoral Context 29 Not identified by the Danish studies, but nonetheless highly influential as a news factor driving press image coverage in New Zealand in 2014 was the self-interest of the press. Other international studies have found that decisions about what is newsworthy often reflect the media’s own self-interest; that they frequently give disproportionate attention to issues that correspond to the agendas of their own profession, particularly when they feel like politicians are challenging themfor control of the news agenda (Iyengar 2005; Tresch 2009). Dirty Politics was one of those issues. The New Zealand press would no doubt argue that the issue was as much in the public interest as its own interest. However, what distinguishes this as being a press interest issue foremost is the disproportionate levels of coverage of Key in number and area in relation to every other event that had taken place during the whole of election year, the negative tonal coverage which was out of balance with the rest of the year and the rest of the 2014 campaign, and the consistency of coverage levels with the ‘Teagate’ phase of the 2011 election campaign, which at its heart was also a disagreement between the press and the Prime Minister over control of information. Internal party polling by both National and Labour confirmed that Dirty Politics was widely assessed as being of little relevance to the general voting public and made no difference to the overall levels of voter support (Joyce 2015, Mills 2015).

Dirty Politics represented ten days of the 30 day official election campaign when it was in the interests of the public to be seeing other party leaders, not just John Key, in press coverage. Instead what they were faced with was a very unbalanced campaign media environment. The challenger and minor leaders were relegated as bystanders to the main event. Half of the minor leaders had their images published twice or less during this phase. Given what we know about time of voting decision discussed earlier, and polls opening for advance voting on 3 September, the last thing that minor parties needed in the first ten days of the campaign was for their leaders to receive little to no visibility in the press.

It is beyond the bounds of this study to measure the electoral damage caused to those parties that were effectively frozen out of this period of critical press exposure, but being out of sight and not top of mind during that time will not have helped them. Similarly, the study is not able to measure the relationship between Key’s high levels of image coverage and National’s high levels of party vote support. But it certainly will not have harmed National. National was the highest polling party in the 2014 election, with a 47.04% party vote virtually unchanged from 2011, and not needing the support of New Zealand First or the Conservatives to form the next government.

Claire Robinson Visual Bias in a Multi-party Electoral Context 30 Because National’s party vote held up so well it was the minor incumbent support/ blackmail parties, whose leaders were amongst the least visually important during the year, that National needed to help it form the next coalition government. Relevant to the day-to-day workings of the government, but on a treadmill of media irrelevance in the context of the news factors judged by the press to be important to campaign image coverage, these parties are unlikely to grow further in public support without increased visibility in the media. We are left wondering what the electoral landscape would look like now if the leaders of those parties had received the levels of attention that the media anointed kingmaker parties received.

Dirty Politics catapulted the press out of its supposedly neutral role in reporting on the election campaign, and transformed it into a self-interested political actor. When the author’s 2011 analysis exposed the visual press bias in favour of John Key it was painless for the press to critique the methodology and dismiss the findings of a single study. The 2014 evidence confirms that there are deeper issues that need to be addressed regarding the ability of the press to present a fair and balanced picture of the electoral landscape in a multi-party election contest, and especially how to guard against imbalance when the press moves from objective spectator to involved actor in a political incident. This gives rise to the question: If the press was to give visual coverage to political leaders in an election campaign that reflected the MMP electoral system’s original intent to promote greater fairness between political parties and provide more effective representation of minority parties, while meeting the press’ own high standards of fairness and balance, what would a more appropriate measure of balance be?

Firstly, must come awareness. With recognition of unconscious bias a precondition for behaviour change, the press needs to recognise the situations under which it is likely to be visually biased9, such as when it feels that politicians are challenging it for control of the news agenda, and then develop safeguards to prevent it from impacting on fair and balanced electoral coverage. In terms of those safeguards, adherence to a narrow range of news factors is clearly not the answer, and proportionality based on previous election results or opinion poll standing is difficult. This study found that National, Labour and the Greens all received less proportional coverage than their 2011 election results, Feb/March 2014 public polling results, and their 2014 actual election results would have apportioned them. All the other parties received more. The three largest parties might have reason to argue they should have had more press image coverage to remain proportionate. However, given the already large lead John Key had over all his competitors, it is doubtful that

Claire Robinson Visual Bias in a Multi-party Electoral Context 31 newspaper readers or opposition parties would be asking for yet more coverage to be allocated to the incumbent leader just to keep him in proportion with opinion poll and election results. In a similar vein, it is doubtful that minor parties with lower opinion poll and party vote support than newspaper coverage would be asking for less coverage of their leaders in order to maintain proportionality. As a measure of fairness, proportionality may not be the most appropriate or reliable instrument.

At the very least there should be agreement about a minimum level of visual coverage each leader might reasonably expect to receive throughout an election campaign. The inevitable unforeseen issues that distract the attention of the press at any time in the campaign may still be covered, but at least the parties not immediately involved can still expect a minimum level of coverage at the same time. This is important. If by the image selections they make the press effectively support the electoral fortunes the incumbent and a few newsworthy parties we may question the independence of the press from political power, the ability of voters to make well informed and objective voting choices and the ability of political parties and candidates to participate in a genuinely fair contest in which all have the opportunity to be seen and heard.

Finally a word about the differences between the antipodean and the northern hemisphere research. This study has not been possible to find a structural bias effect across multi-party electoral contexts. The difficultyomparing c results here is that the Danish studies are longitudinal and not specifically visual in focus, and this one compares two elections in part, and new findings from a single election study in another. At the same time, the different findings highlight the importance of studying bias in multi-party contexts all over the world, especially visual bias which may otherwise go unchecked, if we are to gain a more internationally comprehensive understanding of the issue and find solutions to the problem of fairness and balance in multi-party election coverage.

Acknowledgements

The author would like to acknowledge the assistance of Gillian Ransom and Anna Robinson in the collection and coding of the image data.

Claire Robinson Visual Bias in a Multi-party Electoral Context 32 NOTES

1. http://www.electionresults.govt.nz

2. Advance voting opened on 3 September 2014. Increasing numbers of New Zealand voters are choosing to make an advance vote. In 2014 717,579 voters made an advance vote, up from 334,558 in 2011 and 270,427 in 2008

3. New Zealand Herald (7 November 2011: A8) Letters to editor

4. New Zealand Press Council Statement of Principles. http://www. presscouncil.org.nz/principles.php, accessed 20 December 2014

5. Armstrong, John. (2012, 29 November). “Allegation of election coverage bias doesn’t wash.” The New Zealand Herald. Retrieved from http:// www.nzherald.co.nz/nz/news/article.cfm?c_id=1&objectid=10850663

6. New Zealand Election Study data retrieved from http://www.elections. org.nz

7. http://find.ipsos.co.nz/Fairfax-Ipsos/14.02/Poll14.02.15/index. html; and http://www.nzherald.co.nz/nz/news/article.cfm?c_ id=1&objectid=11221487, accessed 20 December 2014

8. http://www.electionresults.govt.nz

9. “How to lean against your biases: A conversation with Daniel Kahneman” American Press Institute. http://www.americanpressinstitute.org/ publications/good-questions/lean-biases-conversation-daniel- kahneman/, accessed 1 March 2015

Claire Robinson Visual Bias in a Multi-party Electoral Context 33 REFERENCES

Adatto, Kiku. 2008. Picture Perfect: Life in the Age of the Photo Op Princeton: Princeton University Press.

Adler, R.B., Rosenfeld, L.B. and Proctor, R.F. 2007. Interplay: The Process of Interpersonal Communication New York: Oxford University Press.

Ballew, Charles C., & Todorov, Alexander. 2007. “Predicting political elections from rapid and unreflective face movements.” Proceedings of the National Academy of Sciences of the United States of America 104(46): 17948-17953.

Banducci, S. 2002. “Gender and Leadership.” In Proportional Representation on Trial eds. Jack Vowles, Peter Aimer, Jeffrey Karp, Susan Banducci, Raymond Miller and Ann Sullivan. Auckland: Auckland University Press.

Bean, C. 1992. “Party Leaders and Local Candidates.” In Electoral Behaviour in New Zealand ed. M. Holland. Auckland: Oxford University Press.

Bean C. and Mughan, A. 1989. “Leadership Effects in Parliamentary Elections in Australia and Britain.” Amercian Political Science Review 83: 1165-1179.

Bucy, Erik P, and Grabe, Maria Elizabeth. 2008. ““Happy warriors” revisited: Hedonic and agonic display repertoires of presidential candidates on the evening news.” Politics and the Life Sciences27(1): 78-98.

Capelos, Tereza. 2010. “Feeling the Issue: How Citizens’ Affective Reactions and Leadership Perceptions Shape Policy Evaluations.” Journal of Political Marketing 9(1): 9-33.

Caple, Helen, and Bednarek, Monika. 2013. Delving into the Discourse: Approaches to News Values in Journalism Studies and Beyond University of Oxford: Reuters Institute for the Study of Journalism.

D’Alessio, Dave, and Allen, Mike. 2000. “Media Bias in Presidential Elections: A Meta Analysis.” Journal of Communication, Autumn, 133 - 156.

Dalen, Arjen van. 2012. “Structural Bias in Cross-National Perspective: How Political Systems and Journalism Cultures Influence Government Dominance in the News” The International Journal of Press/Politics 17(1): 32-55.

Graber, Doris A. 1984. Processing the News: How People Tame the Information Tide New York: Longman Inc.

Claire Robinson Visual Bias in a Multi-party Electoral Context 34 Green-Pedersen, Christoffer, B.Mortensen, Peter, and Thesen, Gunnar. 2013. Who gets in the news? A news value and political context perspective on media appearances of political parties Paper presented at the ECPR General Conference Bordeaux, France. http://ecpr.eu/Filestore/PaperProposal/d83ccc69-70fe-4bbb- bff5-c359b5cde2dc.pdf

Green-Pedersen, Christoffer, B.Mortensen, Peter, and Thesen, Gunnar. 2015. “The Incumbency Bonus Revisited: Causes and Consequences of Media Dominance.” British Journal of Political Science August 2015: 1-18.

Groeling, Tim. 2013. “Media Bias by the Numbers: Challenges and Opportunities in the Empirical Study of Partisan News.” Annual Review of Political Science 16: 129- 151.

Hager, Nicky. 2014. Dirty Politics: How attack politics is poisoning New Zealand’s political environment Nelson: Craig Potton Publishing.

Hall, Crystal C; Goren, Amir; Chaiken, Shelly; & Todorov, Alexander. 2009. “Shallow Cues with Deep Effects: Trait Judgments from Faces and Voting Decisions.” In The Political Psychology of Citizenship ed. E Borgida, C. M. Frederico & J. L. Sullivan. New York: Oxford University Press.

Hopmann, David Nicolas, Vliegenthart, Rens, Vreese, Claes H. de, and Albaek, Erik. 2010. “Effects of Election News Coverage: How Visibility and Tone Influence Party Choice.” Political Communication 27 (4): 389-405.

Hopmann, David Nicolas, Vreese, Claes H. de, and Albaek, Erik. 2011. “Incumbency Bonus in Election News Coverage Explained: The Logics of Political Power and the Media Market.” Journal of Communication 61: 264-282.

Hopmann, David Nicolas, Aelst, Peter van, and Legnante, Guido. 2012a. “Political balance in the news: A review of concepts, operationalizations and key findings.” Journalism – Theory, Practice and Criticism 13(2): 240–257.

Hopmann, David N., Elmund-Praestekaer, Christian, Albaek, Erik, Vliegenthart, Rens, and Vreese, Claes H. de. 2012b. “Party media agenda-setting: How parties influence election news coverage.” Party Politics 18(2): 173-191.

Iyengar, Shanto. 2005. “Speaking of Values: The Framing of American Politics.” The Forum 3: 2005 Article 7

Iyengar, S., and D. R. Kinder. 1987. News That Matters: Television and American Opinion Chicago: University of Chicago Press.

Claire Robinson Visual Bias in a Multi-party Electoral Context 35 Jenkins, Richard W. 1999. “How Much is Too Much? Media Attention and Popular Support for an Insurgent Party.” Political Communication 16(4): 429-445.

Joyce, Steven. 2015. “The 2014 Campaign: Working for New Zealand.” In Moments of Truth: The New Zealand General Election of 2014 eds. Jon Johansson and Stephen Levine. Wellington: VUW Press

Kahneman, Daniel. 2011. Thinking Fast and Slow Farrar, Straus and Giroux: New York

Leathers, D. and Eaves, M. H. 2008. Successful Nonverbal Communication: Principles and Applications Boston: Pearson.

Lenz G. S. and Lawson, C. 2011. “Looking the Part: Television Leads Less Informed Citizens to Vote Based on Candidates’ Appearance.” American Journal of Political Science 55 (3): 574–589.

McAllister, I. and Bean, C. 2006. “Leaders, the Economy or Iraq? Explaining Voting in the 2004 Australian Election.”Australian Journal of Politics and History 52: 604-620.

Marland, Alex. 2012. “Political Photography, Journalism, and Framing in the Digital Age: The Management of Visual Media by the Prime Minister of Canada.” The International Journal of Press/Politics 17(2): 214-233.

Miller, A. H., Wattenburg, M. P. and Malanchuk, O. 1986. “Schematic Assessments of Presidential Candidates.” Amercian Political Science Review 80: 521-540.

Mills, Stephen. 2015. “Survey findings and the 2014 election.” InMoments of Truth: The New Zealand General Election of 2014 eds. Jon Johansson and Stephen Levine. Wellington: VUW Press

Olivola, Christopher Y., and Todorov, Alexander. 2010. “Elected in 100 milliseconds: Appearance-Based Trait Inferences and Voting.” Journal of Nonverbal Behavior 34: 83-110.

Perloff, Richard M. 2014. The Dynamics of Political Communication New York: Routledge.

Potter, Mary C., Wyble, Brad., Hagmann, Carl Erick, and McCourt, Emily S. 2014. “Detecting meaning in RSVP at 13 ms per picture.” Attention, Perception, & Psychophysics 76(2): 270-279.

Remland, M. S. 2nd edition. 2004. Nonverbal Communication in Everyday Life Boston: Houghton Mifflin Company.

Claire Robinson Visual Bias in a Multi-party Electoral Context 36 Riggio, Heidi R., and Riggio, Ronald E. 2010. “Appearance-Based Trait Inferences and Voting: Evolutionary Roots and Implications for Leadership.” Journal of Nonverbal Behavior 34: 119-125.

Robinson, C. 2012a. “Interacting leaders” in Routledge Handbook of Political Mar- keting ed. J Lees-Marshment. London: Routledge: 257-270

Robinson, C. 2012b. “The eyes have it: visual bias in election campaign coverage” in Kicking the Tyres: The New Zealand General Election and Electoral Referendum of 2011 eds. J. Johannson and S. Levine. Wellington: Victoria University Press

Schudson, Michael. 2001. “The objectivity norm in American journalism.” Journal- ism 2(2): 149-170

Stewart, Patrick A., Salter, Frank K., and Mehu, Marc. 2009. “Taking leaders at face value.” Politics and the Life Sciences 28(1): 48-74. Surawski, Melissa K., and Ossof, Elizabeth P. 2006. “The Effects of Physical and Vocal Attractiveness on Impression Formation of Politicians.” Current Psychology 25(1): 15-27.

Thompson, John B. 2011. “Shifting Boundaries of Public and Prinate Life”. Theory, Culture & Society 28(4):49-70

Tresch, Anke. 2009. “Politicians in the Media: Determinants of Legislators’ Presence and Prominence in Swiss Newspapers.” The International Journal of Press/Politics 14(1): 67-90.

Tversky, Amos, & Kahneman, Daniel. 1974. “Judgment under Uncertainty: Heuristics and Biases.” Science 185(4157): 1124-1131.

Waldman, Paul, and Devitt, James. 1998. “Newspaper Photographs and the 1996 Presidential Election: The Question of Bias.” Journalism and Mass Communication Quarterly 75(2): 302-311.

Claire Robinson Visual Bias in a Multi-party Electoral Context 37